THERMAL ANALYSIS: Monitoring Flammability
Many potential sources of ignition can trigger a fire or explosion, including electrical sparks; friction; static discharge; hot surfaces; air streams; and direct-fired burners in ovens, kilns and thermal oxidizers. Since there's no guarantee that an environment will remain completely free of air or a source of ignition, the most reliable means of preventing a fire or explosion is to measure the amount of flammable vapors continuously in a process and limit them to a safe level.
What's Safe?Each flammable substance has a level of concentration in air that is usually expressed as a percent by volume, which is known as its lower flammable limit (LFL) or lower explosive limit (LEL). Below a substance's LFL, the mixture of fuel and air is too lean to support combustion. As the amount of fuel increases, the mixture will eventually become too rich to burn-there will be too much fuel and not enough air. This concentration is known as the upper flammable limit (UFL) or upper explosive limit (UEL).
Between the LFL and the UFL lies the flammable range, where, given a source of ignition, the mixture will readily ignite. Almost all safety authorities require a 4:1 margin of safety below the LFL, based on worst-case conditions. Enough dilution air must always be used to maintain a concentration of less than 25% of the LFL, according to the National Fire Protection Association's (NFPA) standard NFPA 86.
However, a process oven or oxidizer is allowed to operate with only a 2:1 safety margin (up to 50% of the LFL) when continuous flammability analyzers are used. The requirements stipulate real-time, fast-response, continuous analyzers connected so they trigger corrective action at predetermined alarm points. Unprotected processes, which might normally run at only 10 or 12% of the LFL to avoid reaching 25% in case of accidental upset, can operate at much higher vapor concentrations when LFL analyzers are used. The resulting cost savings due to reduced ventilation or increased throughput can be considerable.
Because LFLs are determined empirically, the values published by different authorities vary at different times. These variations may be due to small inaccuracies in test procedures such as sample preparation, vessel size, air flow, temperature, observation methods and monitoring instruments. Measurement results are usually rounded to the nearest 0.1% by volume.
A comparison of the LFL value for 50 common solvents published by six different authorities shows that one-third of the solvents compared disagree by a standard deviation of more than 10%. Given these differences, the LFL of a substance is certainly not an absolute value, and safety codes should be followed to ensure safe operation.
Most published LFL values are calculated at room temperature. As a given mixture is heated, however, its flammability increases so that the concentration required to achieve 100% of its LFL is less. This often-overlooked source of increased danger is illustrated in Figure 1. The yellow band indicates 100% of the LFL of a sample mixture, given the degree of uncertainty in published values as previously mentioned. The red area of danger increases as the temperature increases, while the green "safe" area is reduced. What was safe at 75ºF becomes dangerous at 200 or 400ºF.
Sensor ApplicationsAlthough several different types of sensors are employed as LFL monitors, each is best suited to specific applications. Fires and explosions in what was thought to have been "protected" equipment can occur without warning when a sensor is not capable of doing the job it has been assigned. This is most often caused by a misunderstanding of the different available technologies.
Catalytic-bead sensors are constructed of two small wire coils covered with a catalyst (see Figure 2). One coil is "active," while the other is rendered inert and acts as a reference. A flow of electrical current through the internal coils heats the catalytic coating to a temperature at which the active coil will react with many flammable vapors and gases. This reaction occurs in the form of surface combustion, which, in turn, causes an increase in the sensor's temperature. The resulting electrical change is measured on a companion monitor instrument.
The intensity of the catalytic reaction varies for different flammable substances and with their concentration, so the sensor needs to be calibrated for use with a specific substance or group of closely related substances. Like most reactive compounds, the response of the catalyst will change over time. How quickly it changes depends on the particular flammable substance to which it has been exposed and on the concentration of that substance. Calibration correction might be required for each different solvent or solvent mixture used in the process. Catalytic sensors are also susceptible to poisoning, coating and etching caused by compounds in the sample.
Since a hot surface itself can act as a source of ignition, catalytic-bead sensors are shielded by a sintered metal cap or a series of fine-mesh wire screens. These flame arrestors slow the ability of a flammable gas to reach the catalytic sensing element. Depending on the application, that slow response alone may eliminate a catalytic sensor from consideration as a process monitoring device. The fine passages of the flame arrestors are also susceptible to clogging by condensate and particulate matter, blinding the sensor.
Catalytic sensors can exhibit considerable response drift when exposed to constant levels of flammable vapors. There is no way to ensure that they are responding accurately-or even responding at all-other than by injecting a known concentration of test gas through the flame arrestor. This calibration test should be done frequently, as this type of sensor gives no notice of failure.
Because of these characteristics, catalytic sensors are typically used only as leak detectors. They are generally well suited to area monitoring applications where response times of 10-20 seconds are acceptable; where the atmosphere does not contain condensate, dirt or dust; and where the atmosphere is normally free of flammable vapors.
Combustible gases absorb infrared radiation at certain characteristic wavelengths. A typical non-dispersive infrared (NDIR) detector passes a source of infrared energy through the sample and measures the energy received by one of two detectors (see Figure 3). The active detector responds to wavelengths in the same band as the sample gas, while the other detector measures a reference to compensate for changes within the instrument.
When specific combustible gases are present, they absorb some of the infrared energy and produce a signal in the active detector relative to the reference detector. Energy absorbed by the combustible gas for a given wavelength varies exponentially with the particular gas' absorptivity, the concentration and the path length. This means that infrared detectors must be specifically calibrated for a particular gas and can have very high variations in response factors and linearity for other gases.
Infrared detectors are usually limited to detecting a single combustible gas. Like catalytic-bead sensors, they are best suited to area monitoring applications.
Flame ionization is a well-established measurement technique. A well-designed flame ionization detector (FID) completely burns the sample by pre-mixing a small quantity of sample with hydrogen fuel and injecting it into a burner (see Figure 4). Combustible gases burned in a hydrogen flame produce ions that can be measured as a weak current through an imposed electrical field. The resulting electrical signal is proportional to the amount of carbon present in the sample.
Some combustible gases contain oxygen, halogens or other electro-negative species, which, in general, tend to inhibit the formation of ions. Because of its principle of operation, the FID is often referred to as a carbon-counter.
The response time of this detector is very fast, but its measurement of flammability is indirect. To measure flammability, the substance being sampled must be known. The number of carbon ions in the sample must be converted to the number of specific molecules for that substance, and that concentration must then be expressed as a percentage of the LFL. While this technique works well for single substances, an FID's wide variation in response factors can sometimes make it impossible to accurately measure the flammability of other substances or of mixtures.
A flame temperature analyzer measures the amount of heat given off by a pilot flame as it burns in an explosion-proof measuring chamber (see Figure 5). The small, well-regulated flame heats the tip of a temperature sensor suspended directly above it. The signal produced by the sensor when no flammable vapors are present drives the LFL indicator up to 0% LFL. This failsafe technique is known as a "live" zero, because a weakening or loss of flame caused by lack of fuel will generate a downscale malfunction alarm.
When a flammable sample is drawn into the measurement chamber, it is seen by the pilot flame as an additional source of fuel. This causes the temperature in the area of the pilot flame to increase. Since the meter knows that the increased temperature can only be caused by added fuel (the sample), it rises above zero in direct proportion to the flammability of the sample.
The dynamics of the flame temperature analyzer give it highly uniform response factors for a wide variety of combustible gases.
Response AccuracyFew analyzers react the same way to all substances. A catalytic-bead sensor calibrated to accurately read methane will be wildly inaccurate when exposed to hydrogen gas, while an FID calibrated for heptane might not measure alcohol accurately. Both of these sensors are broad-banded and will react indiscriminately with most flammable vapors to which they are exposed.
An infrared sensor is a narrow-band instrument. It can usually discriminate between the substance of interest and background gases, but it does not respond to gases outside of its narrow range of vision. These variations in infrared sensor response might not pose a significant problem when measuring an atmosphere containing a single substance, but when asked to measure a mixture of different vapors, catalytic, FID and infrared instruments usually fail miserably.
Flame temperature analyzers react accurately to most flammable substances and usually measure both single substances and mixtures with the same high degree of accuracy. Unlike some sensors, flame temperature analyzers were developed for one specific purpose-to directly measure flammability.
Response SpeedWhile accuracy is certainly a prime concern when selecting a flammability analyzer, consideration must also be given to how quickly a process might get out of control. The analyzer must be able to produce an alarm and initiate corrective actions in time to prevent a disaster. An analyzer with a stated response time of 10 seconds T63 means that the analyzer will take 10 seconds to reach 63% of its final reading. An analyzer with a stated response time of 13 seconds T90 may actually be much faster than the one rated 10 seconds T63.
It's also important to calculate the time it will take for the sample to reach the analyzer. An analyzer designed to mount directly on a process wall or duct may have a sample delivery delay of only a fraction of a second, while a remotely mounted analyzer with a 50-ft sample line could have a sample delivery delay of 10 seconds or more. Add this delay to the analyzer's response time, and the total time to trigger an alarm may be unacceptable.
Figure 6 illustrates an example where the combined sample line delay and the response time of the analyzer produce a 15- second total delay. In other words, the response of the analyzer is 15 seconds behind the real event. By the time the analyzer activates its warning alarm set at 35% LFL, the true concentration in the process is already above 130% LFL.
It's important to consider the corrective action that the analyzer's alarm relays will control. How long will it take for that damper, fan or e-stop to have an effect on the process? Only after you have calculated the reaction time of all these elements will you know the true response time of the system.
Sample ConditioningTo ensure that the vapor concentration reaching the analyzer is exactly the same as the concentration at the point of origin, the sample must always be kept in a vapor state. The analyzer, the sample line and any other elements of the sample train may have to be heated to keep the sample above its dew point. The sample will condense if it's allowed to cool, which causes drop-out and possible clogging. Additionally, light condensation will coat the walls of the sample line, reducing the concentration that reaches the analyzer and causing false readings. Heavier condensation might clog the sample line, causing the analyzer to shut down the process.
When determining the operating temperature required for the analyzer and sample line, it's important to calculate the flash point or condensation temperature of all substances in the sample atmosphere, even those that are not flammable. Any substance can clog the sample line or analyzer if it is allowed to condense.
Smart MonitoringUsing the correct monitoring system will not only prevent personnel injury and property damage but also improve process efficiency and productivity while working within the required safety codes. When selecting a flammability analyzer, it's always a good idea to enlist the advice of a specialist in the field.
Do not assume that "one size fits all" or that the analyzer that was correct for a previous job will also be the right choice for another application. Whether new or similar to an existing process, the specific details of each application need to be examined closely to prevent disaster. The instrument of choice might vary from one application to the next, but the correct process flammability analyzer will always be fast, accurate and fail-safe.
For more information about flammability monitoring, contact Control Instruments Corp. at 25 Law Dr., Fairfield, NJ 07004; (973) 575-9114; fax (973) 575-0013; e-mail firstname.lastname@example.org ; or visit http://www.controlinstruments.com .
SIDEBAR: Application ExampleConsider an oxidizer that might have to destroy a VOC stream from one or more sources, such as a tunnel or batch kiln. Danger is present when the inlet stream suddenly gets rich enough to ignite or explode. As an example, the exhaust streams of most modern process curing lines tend to be highly concentrated and can run out of control, raising the VOC concentration in the vessel and associated ducts above 100% of the LFL.
In this typical installation, LFL monitors are used on the exhaust ducts of various sources, allowing them to operate at high vapor concentrations while protecting the system from fire and explosion. An LFL analyzer on the inlet duct to the oxidizer protects it from dangerous concentrations produced by rich vapor streams, and an FID at the outlet of the oxidizer ensures that it does not exceed emission levels into the atmosphere. Depending on the type of operating permit, the use of an FID may also allow the oxidizer's operating temperatures to be reduced to more economical levels.